The Beginning of the End for Terraform?(medium.com) |
The Beginning of the End for Terraform?(medium.com) |
Vagrant, Terraform, Vault, Consul and Nomad were just too good. HashiCorp's biggest competitor against Terraform and Vault Enterprise was their own stuff! If your platform team was even moderately competent, you had no need for any of that stuff, especially at the prices they were asking for!
Personally, I was hoping for Microsoft to acquire them.
Terraform as a replacement for ARM is _literally_ the dream, and Vault would have been a really good underlying service for Azure Key Vault.
Now that I think about it, I'd love a universe in which GitHub is Azure DevOps, TFE-but-better is Azure RM, and Vault is Azure Key Vault. Alas, that is not reality.
I want to think that Microsoft made a bid but lost out to IBM.
All that said, OpenTofu is a drop-in replacement for Terraform; nearly zero work required. I haven't tried OpenBao yet, but I presume that it's the same situation there. There's enough steam propelling them to make me think that they'll pull through for the long haul.
no, they removed a lot of things, espeically on the storage side.
I would rather expect MS to declare support for OpenTofu. They don't even have to be involved in it directly.
The main work is ensuring azure providers compatibility (which is formally maintained by hashi). Add compatibility with Tofu to Azure Verified Modules [0] (a great project btw), and we have a clean way out of this by providing real help where it's most needed.
Because at less than 2% market share they are a beneficiary.
The main reason orgs move to multi-cloud is they have to support another cloud for business reasons - AI, cost optimization, etc.
When they replatform internal ops to Terraform any cloud provider can become a viable option. Beforehand, not at all.
Why IBM might not be successful with this acquisition, it is likely the only way they could really grow market share.
> This leads me to my first concern, IBM has a conflict of interest. IBM has a cloud offering, admittedly with a 1.8% market share. Why would they want to keep developing tools that frankly benefit their competitors more than them?
Because they'll make bank off of Hashicorp Cloud Platform.
> Another option is to ‘go native,’ with each of the big three hyperscalers offering its dedicated options: Cloud Formation with AWS, Bicep with Azure, and Config Controller in Google Cloud.
Not an option. People manage way, way more than just AWS, Azure and GCP with Terraform. GitHub, Cloudflare, Tailscale, etc ...
It's an interesting time for Hashistack for sure, but I don't think the industry is prepared to move away from the tooling.
Now GCP, AWS and Azure just sell a superior managed version for cheaper. So they lost their income stream.
Open source products tried to change the license so the big guys would at least have to pay them something to make money on their work.
Better stated, of course, but you get the point.
The open source license is designed to guarantee your liberties. The moment you start to remove some liberties you have neutered it.
Meta's LlaMa for example is "OSS" but under a custom license which effectively prohibits hyperscalers from using it
The real reason might be VCs. The companies going closed source have raised from VCs and they are pushed to more and more growth just changing licenses and taking control.
https://www.jeffgeerling.com/blog/2024/corporate-open-source...
If you prefer video, this is actually a transcription of the youtube video here :
Certainly one of the reasons why open source is popular in software is that it gives you options for maintenance prices. You can do it yourself or pay any number of consulting agencies to do it for you, or the creators of the software. When it becomes locked to one vendor suddenly the market economics change hugely. Now that vendor can crank the price up extremely high, basically until just before the point your willing to engage in a hugely expensive software refactor to move to a different product.
Open standards were supposed to help make it easier to move to new products, but in reality, it rarely is that clean. E.g. look at SQL, while it's easier maybe to move from one SQL database to another than from one completely custom database to another, it's still a massive amount of work due to details of each SQL server.